According to my Radio Shack SPL meter, I can hit at least 127 decibels — which is where the meter tops out. Now, I have not yet replicated that process in a more controlled test, in a suitably calibrated environment. But I would be interested to do so.
Here's Jaime Vendera (measured at 120 db) doing it: https://youtu.be/QGk8nXs6Aao Apparently unless you're at the resonant frequency you need over 163 db to shatter glass: http://decibelcar.com/menugeneric/87.html
> "Even if you could excite the cup, that doesn't guarantee it would break. Fracture depends on the size of the initial defects."
I have to wonder if Jaime tapping the glass with a spoon in that video not only reveals the frequency he needs, but more crucially introduces tiny fractures.